Shekhar U. Kadam
a,
N. N. Misra
*b and
Nobuhiro Zaima
c
aResearch, BioAtlantis Ltd., Tralee, Co., Kerry, Ireland
bGTECH, Research & Development, General Mills India Pvt. Ltd., Mumbai, India. E-mail: misra.cftri@gmail.com; Tel: +91 22 420 31566
cDepartment of Applied Biological Chemistry, Graduate School of Agricultural Science, Kindai University, 204-3327 Nakamachi, Nara City, Nara 631-8505, Japan
First published on 11th March 2016
Chemical imaging based on mass spectrometry is an emerging technology which has opened opportunities for fundamental research in food science. The ability to quantitatively determine the spatial distribution of several chemical entities simultaneously makes it a method of choice for chemical characterization of plant materials and foods. In this review, an overview of the ionization methods is presented, followed by a discussion of the current state of the art in mass spectrometry imaging (MSI) of relevance to practices in food research. The known applications, analytical challenges, and future research potential are highlighted. MSI has been successfully utilized to a variable degree for obtaining information about spatial distribution of a variety of molecules and metabolites in foods. It allows visualizing the spatial molecular maps without extraction, purification, separation, or labelling of foods. The method is likely to find wider adoption with developments in ionization sources and inter-laboratory collaborations.
Chemical imaging methods allow identification of differences in chemical composition and structure of biological tissues.6 Spectroscopic techniques such as Fourier-transform infrared (FTIR),7 NIR8 and Raman9 have been frequently used to identify the spatial distribution of bioactive compounds and chemical markers in food matrices. In recent times, however, the interest in use of ambient ionization based mass spectrometry as a tool for chemical analysis and mass spectral imaging of biological matter is steadily growing. This trend is driven by the success of MSI in several other areas of science and engineering, including pharmaceuticals, medicine10 and polymers.11,12 While the use of MS based food-omics is well established, the use of mass spectrometry imaging (hereafter abbreviated as MSI) in food research is sparse. Low spatial resolution, limitations in molecular identification capabilities, low throughput and requirement of very specific instrumentation for desired results render MSI into a perfect tool for direct analysis of biological tissue.
Mass spectrometry based chemical imaging involves combining on a common platform, the elemental or molecular mass measurement obtained via mass spectrometry and spatial information obtained via imaging. This enables visualizing the spatial distribution of elements/molecules of interest on complex surfaces. The overall framework of MSI is similar to near-infrared chemical imaging,13 which has earned recognition for food quality monitoring within the recent past.14 Mass analysis of molecules/isotopes are to MSI, what molecular vibrations and infrared spectrum analysis are to near-infrared chemical imaging. However, MSI certainly offers a very high chemical specificity compared to infrared imaging, especially for complex biological matrices such as foods and plant materials.
MSI involves the general steps of sample preparation, desorption/ionisation of the sample, followed by analysis of ion masses using a mass spectrometer, the image registration and subsequent multivariate analyses. MSI ionization methods allow desorbing analytes directly from the food surface, thereby avoiding traditional liquid extractions. What differs essentially between different MSI techniques is the way ions are generated. The ionization of the sample is an important aspect to be considered for analysis of a given food sample, and is discussed later in Section 2. The mass analysis is carried out using mass spectrometers, which depending on the instrument, could either be a time of flight (TOF), orbitrap, or Fourier transform ion cyclotron resonance (FTICR). The topic of mass analysis has already rendered itself into several volumes of literature.15,16 For ease of presentation, the sample preparation involved is relegated to Section 3 – after describing the ionization methods. In practice, the mass spectrum is recorded as the ionization spot moves spatially, generating data which is automatically mapped to spatial coordinates; this is known as ‘image registration’. The concept of image registration is explained within the Data analysis section.
MSI is an exceedingly multidisciplinary field and its application to biological systems, including food and plant material, often involves the joint efforts of chemists, food scientists, physicists and statisticians.17 There has been a surge of interest in use of MSI as a tool for pathological analysis,18 understanding pharmaceutical mechanisms, drug metabolism19 and biomarker discovery. This is evident from the growing number of publications in this area (see Fig. 1). However, these tools are relatively less known and yet to gain popularity among food scientists. MSI offers several fundamental research opportunities within food science, as will be discussed later in the review. Within its true spirit, MSI has a huge potential to enable a detailed understanding of the biological processes in foods over a range of different length scales, ranging from sub-cellular to tissue, all the way up to whole foods.
![]() | ||
Fig. 1 Number of publications retrieved from Web of Science, for ‘mass spectrometry imaging’ (left) and ‘food mass spectrometry imaging’; accessed on 27 June, 2015. |
In this review, the ionization methods, sample preparation, and data analysis in MSI are briefly introduced, followed by a discussion of the progress made until date in its application to probe food systems. The potential opportunities and challenges in chemical imaging of complex food matrices using MSI have been factually and also speculatively highlighted. We target this review at food scientists and mass spectroscopists, who are unaware of the work from the other field. We hope that this review will attract their interest to adopt this tool in advancing food science.
For the sake of brevity, a discussion of other ionization methods, such as Laser Ablation Electrospray Ionization (LAESI), Probe Electrospray Ionization (PESI), and Laser Electrospray Mass Spectrometry (LEMS) has been omitted here. These are also popular for MSI and have been successfully applied to biological samples.27–31
DESI is based on use of charged droplet flux for ionization and desorption. Similar to MALDI-MSI, DESI is characterized by a lower spatial resolution, but equips an analyst with a tool to analyze small molecules at ambient pressure. On the contrary, SIMS is performed under vacuum conditions. Until recently, MALDI-MSI was ordinarily performed under vacuum; however, Harada et al.32 successfully demonstrated the development and application of an atmospheric pressure laser desorption/ionization microscope (AP-LDI) imager.
Spatial resolutions that can be routinely obtained with MALDI, DESI and SIMS are in the order of 15 μm, 250 μm, and ca. 1 μm respectively. However, with nanoSIMS a lateral resolution of ∼30–50 nm can be achieved.
For MALDI, the sample is typically flash frozen using liquid nitrogen and cryosectioned, following which the chemical matrix is applied. When obtaining reasonable cryosections turns out difficult, the freshly cut surface can be blotted on to pre-coated cellulose membranes for direct loading. This method, known as imprinting or blotting, has been successfully employed for MSI of potato tuber.34 Sometimes, an adhesive tape is useful for mounting samples where the thaw-mounting preparation of the sample section is difficult (e.g. plant seeds or whole-body sections).35,36 When preparing samples for MALDI, inhomogeneous distributions of matrix crystals can result in position dependent ion signals. This could potentially lead to errors and should be taken care of. In recent times, the sensitivity of MSI has been enhanced by coupling it to thin layer chromatography (TLC) blotting. This overcomes the difficulty arising due to competition between numerous molecular species for ionization in non-separated matrices (which permits identification of ‘easy to ionize’ molecules only).37
For SIMS imaging, removal of water and minimisation of salts in the sample is the first requirement. Rapid freezing methods using liquid ethane or propane, which prevent water crystallisation, while keeping the sample frozen and subliming off the vitreous ice are recommended. Imprinting of samples onto hard surfaces prior to SIMS imaging experiments has also been employed to boost signal intensities compared to direct imaging; e.g. leaf samples.38
DESI being an ambient ionization technique in which desorption and ionization occur in the native environment, often minimal or no sample preparation is required. The reduced sample preparation steps compared with vacuum imaging techniques is a specific reason for the rising popularity of DESI. The positioning of the nozzle, the sample, and the inlet of the mass spectrometer is crucial for success with DESI-MSI.39 The use of solvent extraction and/or heating during imprinting for DESI-MSI has also been successfully applied for transfer of chemicals.40
![]() | ||
Fig. 2 A schematic describing the structure of chemical imaging data obtained by mass spectrometry. This exemplary figure shows a MALDI ionisation method. |
The analysis of MSI data typically begins with the application of denoising and smoothing routines, which help to subtract the background noise from the spectrum, followed by optional normalization of the mass intensities. Methods such as nearest neighbour averaging and edge-preserving smoothing algorithms have been proposed for reducing the noise in images.43,44 Wavelets based smoothing and data reduction has also been explored45 and is likely to gain popularity in the near future considering its versatility and speed. Following the application of pre-processing routines, the next challenge involves the feature extraction. The large datasets from MSI studies demand automated methods of data analysis to find differences among spatial distribution in different samples. Many of the assumptions made in the analysis of LC-MS or GC-MS omics datasets are not suitable for imaging.17 For such complex data analysis problems, multivariate statistical methods have stood the test of time, and are also useful for MSI data. Alexandrov et al.41 developed and demonstrated the use of unsupervised cluster analysis, through which m/z images (corresponding to a metabolite) are clustered according to their spatial similarity, resulting in each cluster containing spatially similar metabolite images. Recently, Jaumot and Tauler46 proposed the use of multivariate curve resolution for MSI data, which allows identifying components in distribution maps using resolved pure high-resolution mass spectra. Other options for data analysis include chemometrical methods such as Multivariate Image Analysis (MIA) and Principal Component Analysis (PCA). The recently developed package Cardinal for open source software R allows segmenting and classifying mass spectrometry images, and is based on mixture modeling and regularization statistics.47 Detailed protocols and subtleties of MSI data analysis on commercial software, have been reviewed in a recent book by Setou.48 The general framework for MSI data analysis and necessary precautions have been reviewed by Jones et al.17
MALDI-MSI has been used to visualize the spatial distribution of lipids in rice;49 see Fig. 3. It may be noted from Fig. 3 that the lysophosphatidylcholine (LPC) (18:
2) and LPC (18
:
1) are distributed in the outer area of rice endosperm. On the other hand, LPC (18
:
0) is localized in the core of rice. The distinct distribution of LPC species has been suspected to be a possible reason for the high quality of sake (Japanese traditional alcoholic beverage) made from rice.49 The distribution of triacylglycerols (TAGs) in avocado mesocarp was studied by Horn et al.50 using MALDI-MSI. The mesocarp was found to be rich in monounsaturated fatty acids, and TAG-52
:
2 (TAGs with two oleic and one palmitic acid). However, Horn et al.50 for the first time found that the species with oleic acid at each position, TAG-54
:
3, was relatively inhomogeneous in the mesocarp. This distinct behaviour requires further analysis.
![]() | ||
Fig. 3 Visualization of lipid species in rice grain. Distribution of (a) m/z 520; lysophosphatidylcholine (LPC) (18![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
To avoid matrix-related ions in the spectrum and additional steps in the sample preparation procedure, matrix-free methods such as nano-structure-initiator mass spectrometry (NIMS) and nanowire-assisted laser desorption ionization (NALDI) have also emerged. These methods use an active nanostructured surface to couple the laser energy for the desorption/ionization of analytes present on the surface.51 Essentially NIMS uses a porous, nanostructured silicon substrate doped typically with siloxanes or silanes. NIMS and NALDI methods are emerging techniques in lipid surface analysis by virtue of enhanced sensitivity. SIMS techniques also have the ability to produce higher spatial resolution.
MSI methods can be coupled with additional analytical methods such as thin layer chromatography (TLC) to gain more information at the molecular level. Goto-Inoue et al.52 developed the TLC-Blot-MALDI-MSI technology, discussed earlier (Sample preparation section). TLC-Blot involves transferring separated lipids from TLC plates onto polyvinylidene fluoride membranes to achieve better sensitivity, mass resolution, and minimal background interference in the analysis.53,54 Zaima et al.53 employed this method to visualize and identify major phospholipids, including phosphatidylethanolamine, phosphatidylinositol, phosphatidylserine, phosphatidylcholine and sphingomyelin in tuna fish (Thunnus thynnus).
When food products are tested for their shelf-life, spatial heterogeneity in degradation is often overlooked. Consequently, localised degradation could severely influence end-user perception, while it could remain undetected during holistic evaluations. Within this context, advances in imaging of lipids have enabled researchers to track spatial degradation of lipids directly from meat tissue without extraction.55 Identification of lipid degradation mechanisms directly from muscle tissues will empower food scientists to better understand safety and shelf-life issues related to meat products without any chemical extraction or modification of lipids.
Recently, MSI has been used to find spatial distribution of nutrients and bioactive compounds in foods such as strawberries, celery, sage leaf, potato, cooked chicken egg and turkey.59 Kim et al.60 used MALDI-MSI for visualizing the bioactive strictinin (an ellagitannin found in green tea cultivars), on orally dosed mouse kidney tissue. Authors found intact strictinin within kidney after 1 h of oral dosing and suggested that 1,5-diaminonaphthalene is the best matrix for high sensitivity in the negative ionization mode. Flavonoids in foods have also been subject of MSI studies. Seyer et al.61 used SIMS to localize the flavonoids in fresh pea and Arabidopsis thaliana seeds. Subsequently, Li et al.62 detected and visualized free flavonoids, flavonoid glycosides and saponins with mass difference of 0.02 Da by using tandem mass spectrometry imaging.
Polyphenols, particularly anthocyanins, are known for many beneficial effects to the human health. Using MSI, these compounds have been imaged by Yoshimura et al.63,64 in blueberry and black rice. The authors found that black rice anthocyanins with a pentose moiety were localized in the entire pericarp, whereas anthocyanin species with hexose moiety were focally localized in the dorsal pericarp. On the contrary, in case of blueberry, the anthocyanins delphinidin and petunidin were dominantly distributed in the exocarp, while cyanidin, peonidin and malvidin were found in exocarp as well as the endocarp.
Cabral et al.40 successfully imaged strawberry fruit (Fragaria × ananassa Duch.) and ginkgo leaf (Ginkgo biloba L.) sections by imprinting the samples on TLC plates, followed by DESI-MSI. In the strawberry analysis, authors found that sugar ions were distributed throughout the fruit whereas the anthocyanin pelargonidin-3-glucoside was highly concentrated in the margin region of strawberry. However, the exact location of anthocyanin pelargonidin-3-glucoside should be further investigated as it could be the result of pressure induced during imprinting.
The knowledge of anthocyanins (or other bioactive or macromolecule) distribution patterns in grains could serve to guide and accelerate research in the area of grain dry-fractionation.65,66 One could, in principle, design processes for separation of these regions through dry milling and classification to obtain natural ingredients for food use. Furthermore, extending this concept to the case of plant materials, extraction efficiencies for bioactives can be increased by using selected material fractions rich in the component, against the current practice of whole plant material usage in most cases.
![]() | ||
Fig. 4 Spatial distribution of the natural carbohydrates and lycoperosides F and/or G and/or esculeoside A on cherry tomato slice through ion maps of (a) m/z 218.983, (b) m/z 381.022 and (c) m/z 654.716. Adapted from Nielen and van Beek72 by courtesy of Springer. |
Arsenic (As) is a toxic element entering the food chain through arsenic contaminated water used for irrigation. Rice (Oryza sativa) is a staple food in As-epidemic zones of world. Moore et al.73 employed high-resolution SIMS (nano-SIMS) to probe the distribution of arsenic in rice grains. They observed that arsenic was confined to the sub-aleurone endosperm cells in association with the protein matrix rather than in the aleurone cells.
The adulteration of food products is of primary concern to consumers and regulatory bodies. Melamine is well known adulterant in protein and infant food products. DESI-MSI has been used to detect the adulterant melamine in edible eggs. According to Yang et al.74 significant proportions of melamine are confined to the egg white and only little amount is present in the egg yolk region. On a similar subject, lipid transfer proteins (LTPs) are pan-allergens found in plants, cf. Pru p-3 in peach. Bencivenni et al.75 used MALDI-MSI to confirm the localization of the tomato LTP isoforms in the seeds and its absence in peel and pulp tissues. Thus, separation of seed fraction from tomato would eliminate the risk of having LTP in the diet. On the contrary, peach peel is the most dangerous for allergic people since MALDI-MSI shows Pru p-3 LTP distribution in peel.76
Cabral et al.40 successfully imaged potato sprout (Solanum tuberosum L.) to identify the distribution of glycoalkaloid by imprinting the samples on to TLC plates, followed by DESI MSI. It may be recalled that glycoalkaloids are natural toxins occurring in potato, which impart antimicrobial, fungicidal and insecticidal properties to defend against invading pathogens. This group visualised higher intensities for α-chaconine at m/z 852 and α-solanine at m/z 868 at the sprout tip relative to the remainder of the potato. In addition, all the imprinting conditions showed similar intensity and image quality (see Fig. 5). Based on the aforementioned studies, it may be concluded that MSI is a very useful tool in food research for identifying and localizing allergens, pesticides, toxins, adulterants. This could occasionally help in selective and safe utilization of foods, otherwise known to be harmful and completely rejected.
![]() | ||
Fig. 5 (a) Spatial distribution of potato sprout imprinted directly onto the TLC plate with solvent mixture deposited and heat applied. (a1) The distribution of α-chaconine (m/z 852) and (a2) α-solanine (m/z 868). (b) The mass spectrum was acquired in positive ion mode. MS/MS analysis was also carried out for confirmation. Adapted from Cabral et al.,40 by courtesy of Springer-Verlag. |
Taira et al.78 used MALDI-MSI to identify spatial distribution of metabolites from Panax ginseng and from Capsicum fruits. Ginsenosides are bioactive constituents of Panax ginseng, known for health effects on the immune system, cardiovascular system and central nervous system. MALDI-MSI revealed that the ginsenosides were located in the pericarp and a part of the cortex in the center of the lateral root and the entire fibril of Panax ginseng. The other metabolite, analyzed by Taira et al.78 was capsaicin from Capsicum fruits. It is an alkaloid which was shown to be present at the placenta, pericarp and seed regions of Capsicum fruits. In another distinct study, monoterpenes and 6-gingerol, compounds responsible for pungency and flavor in fresh ginger, were visualized in oil drop-containing organelles of fresh rhizome using atmospheric pressure desorption/ionization based mass-micrometry.32
Burrell et al.34 studied the spatial distribution of metabolites in wheat grain by MALDI-MSI using the matrices such as α-cyano-4-hydroxycinnamic acid (α-CHCA) and 9-aminoacridine. They successfully mapped the grain anatomy to arginine, sucrose and glucose-6-phosphate distributions in the grain. Burrell et al.34 also recorded differences in sucrose distribution amongst wheat grown at different temperatures, with higher temperatures leading to lower yield. This study is of significance considering its broader implications regarding obtaining targeted functionality or selective nutrient enrichment through dry fractionation (as mentioned earlier) or wet extraction routes.
Catabolic processes of leaf senescence and fruit ripening have remained quite under-researched. Müller et al.79 used DESI-MSI to study chlorophyll degradation products (e.g. non-fluorescent chlorophyll catabolites), both directly and after imprinting on PTFE from plant leaf tissues. Imprinting on PTFE enhanced the sensitivity and identification of chlorophyll catabolites concentrated in minute quantities.
Selenium is an important trace element in human nutrition. The distribution of selenium in wheat (Triticum aestivum cv. Hereward) was recently investigated by Moore et al.73 using high-resolution nanoSIMS. The study found that majority of the Se was associated with the protein circumscribing the starch granules of the endosperm and more uniformly present in the aleurone, with some hotspots rich in selenium. Accumulation of Pb, Mg and Cu in sunflower (Helianthus annuus L.) leaves was localized by using LIBS (Laser-Induced Breakdown Spectroscopy) and LA-ICP-MS (Laser Ablation Inductively Coupled Plasma Mass Spectrometry) around close positions of the central vein with resolution of 200 μm.80 Kötschau et al.81 carried out a more detailed study on sunflower leaves where each part of leaf has shown to possess diverse distribution of macro- and microelements. Veins of leaves accumulate most of elements such as Cd, Ce, Cu, La, Mn, while Zn, Fe and S were found to accumulate at the tip of the leaf and K and Ni were enriched in the mesophyll area. Ca, Cr and P were found to be homogeneously distributed over the whole sunflower leaf. Overall, one could appreciate the applications of MSI to not only map organic metabolites, but also mineral constituents.
We opine that in near future ambient pressure MSI will serve as a discovery tool in food research, whereafter it will eventually mature into an in-line/real-time process analytical tool for the food industry.13 This is justifiable considering the rapid advances in MSI instrumentation and the efforts of physicists and chemists to make instruments inexpensive and simpler. In a related context, it may be noted that the diameter of the laser focus (optical resolution) in most MSI instruments is in the order of 100 μm or less, which restricts the spatial resolution of chemical images to ca. 100 μm. However, considerable research efforts have been put to enhance the resolution in space of MALDI-MSI by an order of magnitude (say, 10 μm).82,83 Thanks to the developments in laser physics, high-resolution mass spectrometers with coaxial laser illumination ion sources capable of irradiating areas as small as 40 μm2 (∼7 μm diameter) now exist.84 Moreover, the cost of lasers is also steadily declining.
McDonnell et al.85 noted that MSI studies, at present, are largely individual endeavours, based on their expertise and laboratory infrastructure. In addition, MSI studies are largely focused towards bio-medical applications and consequently often confined to laboratories involved in medical research. Therefore, inter-laboratory collaborations at initial stages could prove useful to food scientists. That said, MSI tools are becoming easily available, though they still tend to be expensive. Development of cheaper ionization sources is expected to partially reduce the cost burden.
This journal is © The Royal Society of Chemistry 2016 |